Data Analytics for Banking Coursework

Author
Affiliation

Pietro Rota

Bayes Business School – City, St George’s University of London

TO RENDER CORRECTLY:

quarto preview “c:/Users/pietr/OneDrive - City St George’s, University of London/Documents/a2025-2026 BAYES/DABF - Data Analytics for Banking & Finance/DABF_code/Groupwork/Groupwork.ipynb” –to html –execute

Introduction

This analysis examines the performance of major companies in the energy sector over the past year, especially looking at the renewable energy segment. This year has been characterised by heightened volatility and structural changes driven largely by geopolitical tensions, inflationary pressures, and shifts in global interest rates. Notably, traditional energy commodities, such as oil and natural gas, have underperformed compared to the previous three years. This reflects a confluence of softer demand and accelerating momentum behind the global transition toward cleaner energy sources.

Show code 
from scipy.stats import jarque_bera
from statsmodels.tsa.stattools import adfuller
from datetime import datetime
from statsmodels.api import OLS, add_constant
from great_tables import GT
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns

print("Current Time:", datetime.now())
Current Time: 2025-12-04 17:51:51.232692

Data Collection

All the Financial information was retrieved from the Bloomberg terminal using the BQL tool for easier date alignment, as these companies are listed on different stock exchanges, they observe different holidays. BQL also allows for currency conversion to British Pounds. The 10 companies that have been chosen are:

Table 1: Companies Table with Bloomberg Industry Classification Standard and Transition Readiness
Company name Bloomberg ticker Col3 BICS sector 4 Transition Readiness
Chevron Corp CVX US Equity New York Integrated Oils Unprepared
First Solar Inc FSLR US Equity NASDAQ GS Renewable Energy Equipment Aligned
Eni SPA ENI IM Equity Borsa Italiana Integrated Oils Developing
Exxon Mobil Corp XOM US Equity New York Integrated Oils Unprepared
Shell PLC SHEL LN Equity London Integrated Oils Unprepared
Vestas Wind Systems A/S VWS DC Equity Copenhagen Renewable Energy Equipment Aligned
Equinor ASA TTE FP Equity Oslo Integrated Oils Developing
TotalEnergies SE EQNR NO Equity EN Paris Integrated Oils Developing
Ameresco Inc AMRC US Equity New York Renewable Energy Project Aligned
NextEra Energy Inc NEE US Equity New York Integrated Electric Utilities Developing
Show code 
def basic_plot(data, NAME):
    fig, ax = plt.subplots(figsize=(12, 6))

    # Plot all columns (companies) against the index (quarters)
    # This is much simpler than the bar chart, as Matplotlib handles the
    # X-axis scaling automatically for line plots on an index.
    data.plot(kind='line', ax=ax)

    # --- 3. CUSTOMIZING THE CHART ---

    # Add titles and labels
    ax.set_xlabel("Quarter", fontsize=12)
    ax.set_ylabel(NAME.upper(), fontsize=12)
    ax.set_title(f"{NAME.capitalize()} Trend of 10 Companies Across 4 Quarters",
        fontsize=14,
        fontweight='bold'
    )

    # Customize the legend: place it outside the plot area
    ax.legend(
        title="Companies",
        bbox_to_anchor=(1.05, 0.5),
        loc='center left',
        borderaxespad=0,
        frameon=False
    )

    # Ensure only the quarter labels are shown on the X-axis
    ax.set_xticks(range(4))
    ax.set_xticklabels(quarters)

    # Add a grid for better readability
    ax.grid(axis='both', linestyle='-', alpha=0.6)

    # Display the plot
    plt.tight_layout()
    plt.show()
def clustered_bar_chart(df, name):
    quarters = ["Q4 2024", "Q1 2025", "Q2 2025", "Q3 2025"]
    plot_df = df.transpose().iloc[-10:]
    plot_df.columns = quarters
    companies = plot_df.index.tolist()

    x = np.arange(len(quarters))  # 4 quarter positions
    width = 0.09                  # small width because 10 companies inside each cluster

    plt.figure(figsize=(14, 7))

    for i, company in enumerate(companies):
        plt.bar(
            x + i*width,                         # shift inside each quarter cluster
            plot_df.loc[company, quarters],           # values for that company
            width,
            label=company,     # show legend only once
        )

    # Fix x-axis tick positions (center the cluster labels)
    plt.xticks(x + width*len(companies)/2, quarters)

    plt.xlabel("Quarter")
    plt.ylabel("Value")
    plt.title(f"Clustered Bar Chart of {name} by Quarter")
    plt.legend(title="Company", loc='center left',bbox_to_anchor=(1.02, 0.5), ncol=1, frameon=False)
    plt.tight_layout()
    plt.show()

def Quick_scatter(Data1, Data2, label1, label2, ylog=False, xlog=False, panelreg = False, groupreg = True):
        # Collect all data for panel regression
    all_data1 = []
    all_data2 = []
    plt.figure(figsize=(12, 6))
    
    if groupreg:
    # ---- classify companies ----
        green_status = {
            "CVX US Equity": "no",
            "FSLR US Equity": "yes",
            "ENI IM Equity": "some",
            "XOM US Equity": "no",
            "SHEL LN Equity": "no",
            "VWS DC Equity": "yes",
            "EQNR NO Equity": "some",
            "TTE FP Equity": "some",
            "AMRC US Equity": "yes",
            "NEE US Equity": "yes"
        }


        # If you want: define green = yes only
        green_companies = [c for c,s in green_status.items() if s == "yes"]
        non_green_companies = [c for c,s in green_status.items() if s == "no"]
        some_companies = [c for c,s in green_status.items() if s == "some"]  # optional


        # Also prepare lists for green/non-green group regressions
        green_x = []
        green_y = []
        nongreen_x = []
        nongreen_y = []
        some_x = []
        some_y = []

    # --- Plot scatter + company-level regressions ---
    for company in company_columns:
        Data1_series = Data1[company]
        Data2_series = Data2[company]

        all_data1.append(Data1_series)
        all_data2.append(Data2_series)
        if groupreg:
            # Add to correct group
            if company in green_companies:
                green_x.append(Data1_series)
                green_y.append(Data2_series)
            elif company in non_green_companies:
                nongreen_x.append(Data1_series)
                nongreen_y.append(Data2_series)
            elif company in some_companies:
                some_x.append(Data1_series)
                some_y.append(Data2_series)

        # Company scatter
        plt.scatter(Data1_series, Data2_series, label=company)

        # Company individual regression
        # sns.regplot(
        #     x=Data1_series.values,
        #     y=Data2_series.values,
        #     scatter=False,
        #     ci=None,
        #     line_kws={'alpha': 0.6, 'linewidth': 1}
        # )

    # --- Convert group lists ---
    if groupreg:
        gx = pd.concat(green_x).values if green_x else None
        gy = pd.concat(green_y).values if green_y else None
        nx = pd.concat(nongreen_x).values if nongreen_x else None
        ny = pd.concat(nongreen_y).values if nongreen_y else None
        sx = pd.concat(some_x).values if some_x else None
        sy = pd.concat(some_y).values if some_y else None

    if panelreg:
        # --- Panel-wide regression ---
        all_x = pd.concat(all_data1).values
        all_y = pd.concat(all_data2).values
        sns.regplot(
            x=all_x, y=all_y,
            scatter=False, ci=None,
            label='Panel-Wide Regression',
            line_kws={'color': 'blue', 'linewidth': 4, 'alpha': 0.2}
        )

    if groupreg:
        # --- Group Regressions ---
        if gx is not None:
            sns.regplot(
                x=gx, y=gy,
                scatter=False, ci=None,
                label='Aligned Companies',
                line_kws={'color': 'green', 'linewidth': 5, 'alpha': 0.2}
            )

        if nx is not None:
            sns.regplot(
                x=nx, y=ny,
                scatter=False, ci=None,
                label='Unprepared Companies',
                line_kws={'color': 'red', 'linewidth': 5, 'alpha': 0.2}
            )

        if sx is not None:
            sns.regplot(
                x=sx, y=sy,
                scatter=False, ci=None,
                label='Developing Companies',
                line_kws={'color': 'gray', 'linewidth': 5, 'alpha': 0.2}
        )

    # --- Formatting ---
    if xlog:
        plt.xscale('log')
    if ylog:
        plt.yscale('log')

    plt.title(f'{label1} vs. {label2} (with group regressions)', fontsize=14)
    plt.xlabel(label1)
    plt.ylabel(label2)
    plt.grid(True, ls="--", alpha=0.6)
    plt.legend(loc='center left', bbox_to_anchor=(1.02, 0.5), frameon=False)
    plt.tight_layout()

def percentage_stacked_bar_chart(df, name):
    quarters = ["Q4 2024", "Q1 2025", "Q2 2025", "Q3 2025"]
    plot_df = df.transpose().iloc[-10:]
    plot_df.columns = quarters
    companies = plot_df.index.tolist()

    # Convert to numeric
    plot_df = plot_df.apply(pd.to_numeric, errors="coerce")

    # Normalize so each column sums to 100

    df_pct = plot_df[quarters].div(plot_df[quarters].sum(axis=0), axis=1) * 100

    x = np.arange(len(quarters))

    plt.figure(figsize=(14,7))

    bottom = np.zeros(len(quarters))

    for company in companies:
        values = df_pct.loc[company, quarters].values.astype(float)

        plt.bar(
            x,
            values,
            bottom=bottom,
            label=company
        )
        bottom += values

    plt.xticks(x, quarters)
    plt.xlabel("Quarter")
    plt.ylabel("Percentage Share (%)")
    plt.title(f"Percentage Stacked Bar Chart of {name}")
    plt.legend(title="Company", loc='center left', bbox_to_anchor=(1.02, 0.5), ncol=1, frameon=False)

    plt.ylim(0, 100)
    plt.tight_layout()
    plt.show()

def simple_bar_chart(data, title, ylabel = "Company"):

    colors = {
    'CVX US Equity': '#1f77b4',
    'FSLR US Equity': '#ff7f0e',
    'ENI IM Equity': '#2ca02c',
    'XOM US Equity': '#d62728',
    'SHEL LN Equity': '#9467bd',
    'VWS DC Equity': '#8c564b',
    'EQNR NO Equity': '#e377c2',
    'TTE FP Equity': '#7f7f7f',
    'AMRC US Equity': '#bcbd22',
    'NEE US Equity': '#17becf'
    }

    plot_data = data.iloc[0, 1:].sort_values(ascending=False)
    bar_colors = [colors[ticker] for ticker in plot_data.index]

    plt.figure(figsize=(10, 6))
    plot_data.plot(kind='barh', color=bar_colors, width=0.9)
    plt.xlabel(title, fontsize=12, fontweight='bold')
    plt.ylabel(ylabel, fontsize=12, fontweight='bold')
    plt.title(f"{title} Analysis", fontsize=14)
    plt.grid(axis='y', alpha=0.3)
    plt.tight_layout()
    plt.show()
Show code 
from pathlib import Path
PATH = Path("C:/Users/pietr/OneDrive - City St George's, University of London/Documents/a2025-2026 BAYES/DABF - Data Analytics for Banking & Finance/DABF_code/Groupwork/Data-Analytics-For-Banking/DF.xlsx")

df = pd.read_excel(PATH, sheet_name="PRICES GBP")
# df['Date'] = pd.to_datetime(df['Date'], unit='D')
df.set_index('Date', inplace=True)
df = df.astype(float)
df2 = df.drop("SPX Index", axis=1)
# Inspect the data

#| label: import_fundamentals
MKT = pd.read_excel(PATH, sheet_name = "MKTCAP")
PE = pd.read_excel(PATH, sheet_name = "PE")
EPS = pd.read_excel(PATH, sheet_name = "EPS")
ROA = pd.read_excel(PATH, sheet_name = "ROA")
ROE = pd.read_excel(PATH, sheet_name = "ROE")
ESG = pd.read_excel(PATH, sheet_name = "ESG")

quarters = PE["Date"]
company_columns = PE.columns[1:]

df = df.dropna()
df.round(4)
CVX US Equity FSLR US Equity ENI IM Equity XOM US Equity SHEL LN Equity VWS DC Equity EQNR NO Equity TTE FP Equity AMRC US Equity NEE US Equity SPX Index
Date
2024-11-01 22.6098 158.5241 11.7630 88.9155 25.790 14.9066 18.1238 48.5307 25.0077 59.8314 4431.3119
2024-11-04 22.9469 163.8469 11.7849 91.5483 25.940 15.3201 18.2802 48.8139 25.8413 60.4893 4409.3007
2024-11-05 22.9860 165.6360 11.8373 91.2690 26.085 13.4186 18.2319 48.8806 27.1597 60.3422 4436.6733
2024-11-06 23.3980 150.5198 11.6541 93.8712 25.925 11.6121 17.4024 47.9095 24.9418 57.8123 4599.7207
2024-11-07 23.3179 151.4242 11.7200 93.2640 25.995 11.9894 17.7310 48.0278 24.3572 57.4442 4598.2294
... ... ... ... ... ... ... ... ... ... ... ...
2025-10-27 26.0762 185.7582 13.9143 86.9507 28.320 14.3080 18.3293 46.9308 31.1609 64.5193 5156.1122
2025-10-28 25.9547 180.4625 13.9548 86.6385 28.330 15.4433 18.2800 46.9788 30.7298 62.9434 5190.0957
2025-10-29 26.6142 182.7537 14.0564 88.0463 28.755 15.6010 18.3470 47.6764 29.9032 61.8176 5209.8821
2025-10-30 26.4350 177.5869 13.9991 87.1968 28.830 15.8986 18.3604 47.1680 29.6282 62.0695 5186.9079
2025-10-31 26.7514 203.2744 13.9917 87.0850 28.475 15.5069 18.1955 47.3439 30.0944 61.9860 5208.8029

237 rows × 11 columns

They were selected to provide a general market overview rather than focusing on a specific jurisdiction. The primary objective is to analyse variations in their business models. The selection includes varied business models: companies that are leading the climate transition like Ameresco, First Solar, and Vestas. Ameresco focuses on integrated clean energy solutions, including cogeneration and hydroelectric systems, First Solar dealing in solar panels, and Vestas creating offshore wind turbines. Then there are companies that are in the middle, such as Eni, Equinor, NextEra, and TotalEnergies, which still largely deal in hydrocarbons, but also articulated transition strategies and plans to at some point switch off their involvement in oil and gas even before the 2050 zero carbon target. Then there are companies like Shell, Chevron and Exxon that have no renewable aims and no intentions to develop them, have opted to focus on exploration, refinement, and sale of hydrocarbons.

Methodology

To analyse the companies in a comparable way, we took the daily returns:

\text{R}_i=\frac{P_{t;i}-P_{t-1;i}}{P_{t-1;i}}

Where P_t is the price today, and P_{t-1} what was the price yesterday. Before this, we dropped all days where any market was closed, totalling 129 lost days for a total of 237 days where we have price data for all firms. Opting to reduce the number of days is the simplest way to make sure that all the data is aligned and there are no empty cells.

Show code 
returns = (df / df.shift(1)-1)
returns = returns.dropna()

SPXreturns = returns["SPX Index"]

full_returns = returns
returns = returns.iloc[:, :-1] # Exclude SPX for initial analysis
Show code 
low = 25
high = 100 - low
q1 = df2.quantile(low/100)
q3 = df2.quantile(high/100)
iqr = q3 - q1

iqr_df = pd.DataFrame({"Names":df2.columns,"Q1": q1, "Q3": q3, "IQR": iqr}).round(4)
iqr_df_sorted = iqr_df.sort_values("IQR", ascending=False)

display(
    GT(iqr_df_sorted)
    .tab_header(title="Per-asset IQR of Prices")
    .opt_stylize(5, color = "gray")
)

# Overall IQR across all prices (flattened, excluding NaNs)
all_vals = df2.values.flatten()
all_vals = all_vals[~np.isnan(all_vals)]
overall_q1 = np.percentile(all_vals, low)
overall_q3 = np.percentile(all_vals, high)
overall_iqr = overall_q3 - overall_q1

print(f"\nOverall IQR across all prices: {overall_iqr:.4f} (Q1={overall_q1:.4f}, Q3={overall_q3:.4f})")
Per-asset IQR of Prices
Names Q1 Q3 IQR
FSLR US Equity 116.1903 152.0578 35.8675
AMRC US Equity 10.9284 20.7633 9.8349
XOM US Equity 81.0785 87.5251 6.4466
NEE US Equity 53.2089 57.475 4.266
CVX US Equity 22.768 25.5948 2.8269
TTE FP Equity 44.6337 47.2574 2.6237
VWS DC Equity 10.9953 13.618 2.6228
ENI IM Equity 11.1933 12.7994 1.606
SHEL LN Equity 25.285 26.83 1.545
EQNR NO Equity 17.9224 19.2164 1.294 

Overall IQR across all prices: 39.1425 (Q1=15.9876, Q3=55.1301)
Show code 
last_prices = df.iloc[-1]
first_prices = df.iloc[0]

(
    GT(pd.DataFrame({"Names":df.columns,"First prices":first_prices, "Last prices":last_prices,"Change":(last_prices-first_prices)}).sort_values(by="Change", ascending=False))
    .tab_header(title="Price Change over the period")
    .opt_stylize(5, color = "gray")
    .fmt(lambda x: f"{x:.2f}£", columns=["First prices", "Last prices", "Change"])
)
Price Change over the period
Names First prices Last prices Change
SPX Index 4431.31£ 5208.80£ 777.49£
FSLR US Equity 158.52£ 203.27£ 44.75£
AMRC US Equity 25.01£ 30.09£ 5.09£
CVX US Equity 22.61£ 26.75£ 4.14£
SHEL LN Equity 25.79£ 28.48£ 2.69£
ENI IM Equity 11.76£ 13.99£ 2.23£
NEE US Equity 59.83£ 61.99£ 2.15£
VWS DC Equity 14.91£ 15.51£ 0.60£
EQNR NO Equity 18.12£ 18.20£ 0.07£
TTE FP Equity 48.53£ 47.34£ -1.19£
XOM US Equity 88.92£ 87.08£ -1.83£
Show code 
(
    GT(pd.DataFrame({"Names":returns.columns, "Sum":returns.sum()}).sort_values(by="Sum", ascending=False))
    .tab_header(title="Sum of Daily Returns")
    .opt_stylize(5, color = "gray")
    .fmt(lambda x: f"{x:.2f}", columns=["Sum"])
)
Sum of Daily Returns
Names Sum
AMRC US Equity 0.63
FSLR US Equity 0.45
CVX US Equity 0.21
VWS DC Equity 0.19
ENI IM Equity 0.19
SHEL LN Equity 0.12
NEE US Equity 0.07
EQNR NO Equity 0.05
XOM US Equity 0.01
TTE FP Equity -0.00

days with the highest returns

::: {#cell-Largest Daily Move .cell execution_count=8}

Show code 
DAYS = pd.DataFrame({"Name":returns.columns, "Negative":returns.idxmin(axis=0), "Positive":returns.idxmax(axis=0), 
#"Difference": returns.idxmax(axis=0)-returns.idxmin(axis=0)
})

(
    GT(DAYS)
    .tab_header(title="Date of Largest Daily Move")
    .opt_stylize(5, color = "gray")
)
Date of Largest Daily Move
Name Negative Positive
CVX US Equity 2025-04-03 2025-04-09
FSLR US Equity 2025-06-17 2025-08-18
ENI IM Equity 2025-04-07 2025-10-23
XOM US Equity 2025-04-10 2025-04-09
SHEL LN Equity 2025-04-04 2025-06-10
VWS DC Equity 2024-11-06 2025-08-18
EQNR NO Equity 2025-06-24 2025-01-02
TTE FP Equity 2025-04-04 2025-10-15
AMRC US Equity 2025-02-28 2025-08-05
NEE US Equity 2025-07-23 2025-04-09

:::

Descriptive Statistics

This table showcases much of the descriptive statistics analysis. Overall, the returns were very low this year, with TotalEnergies going in the negative, renewables showing good daily returns, while the others maintained slim positive averages. Standard deviation varied significantly, with the least risky stock, Eni, having a standard deviation of 1.29% and the most volatile, Ameresco, reaching 6.25%. Coefficient of variation CV=(Standard deviation)/(mean ) serves as a measure of risk-per-unit-of-return. With low means, it becomes harder to interpret; only Eni had a CV under 20, which is the threshold for performing stocks. Ameresco had the strongest mean return coupled with the highest daily volatility, as it is common for the growth renewable plays often look like this. And of note, also Chevron had a lower mean, but also lower volatility. Skewness is varied, ranging from negative, indicating a frequent number of small gains and a few extreme losses, to positive, indicating the potential for larger upside outliers. Renewables showed big upside tails, which are common for high-growth stocks, while all other companies show a negative skewness, meaning that they fall sharply on bad days, rise slowly on good days. Kurtosis shows how much the tails contribute to the total distribution compared to the normal distribution, which signifies the likelihood of extreme outliers. Ameresco with the highest amount, showing a higher probability of extreme returns; the rest of the renewable stocks showed increased probability of extreme returns. Beta (β) systematic risk is derived by using a reference index like the S&P 500, serving as a proxy for the overall economy. Ameresco was the only company with a defensive beta, indicating volatility on par with the index. Other companies showed a beta lower than 1, which indicates greater stability under changing market conditions. Vestas and Equinor had a negative beta, which represents a negative correlation with the index. The 5% daily Gaussian VaR ranges from -2.04% Eni to -10% Ameresco. The 1% VaR confirms substantial downside risk in the volatile renewable energy stocks. However, these estimates are likely biased because the normality assumption is clearly violated for assets with such high kurtosis.

Show code 
from scipy.stats import norm
 
desc_stats = pd.DataFrame({
    "Count": returns.count(),
    "Mean": returns.mean(),
    "Std": returns.std(),
    "CV": returns.std()/returns.mean(),
    "Skewness": returns.skew(),
    "Kurtosis": returns.kurt()
})


index_returns = np.array(SPXreturns)
OLS_returns = returns.reset_index()

betas = {}
for ticker in OLS_returns.columns[1:]:
    y = OLS_returns[ticker]
    X = np.array(add_constant(index_returns))
    model = OLS(y, X).fit()
    betas[ticker] = model.params[1]  # beta coefficient
beta_df = pd.DataFrame.from_dict(betas, orient='index', columns=['Beta'])

ADF_test = {}
ADF_pval = {}
for ticker in OLS_returns.columns[1:]:
    model = adfuller(OLS_returns[ticker])
    ADF_test[ticker] = model[0]  # ADF coefficient
    ADF_pval[ticker] = model[1]  # pval 
ADF_test_df = pd.DataFrame.from_dict(ADF_test, orient='index')
ADF_pval_df = pd.DataFrame.from_dict(ADF_pval, orient='index')

JB_test = {}
JB_pval = {}
for ticker in OLS_returns.columns[1:]:
    model = jarque_bera(OLS_returns[ticker])
    JB_test[ticker] = model[0]  # ADF coefficient
    JB_pval[ticker] = model[1]  # pval 
JB_test_df = pd.DataFrame.from_dict(JB_test, orient='index')
JB_pval_df = pd.DataFrame.from_dict(JB_pval, orient='index')


# VAR 5%
VAR5 = {}
VAR1 = {}
for ticker in OLS_returns.columns[1:]:
    VAR5[ticker] = norm.ppf(0.05, loc=desc_stats["Mean"][ticker], scale=desc_stats["Std"][ticker])
    VAR1[ticker] = norm.ppf(0.01, loc=desc_stats["Mean"][ticker], scale=desc_stats["Std"][ticker])
VAR5_df = pd.DataFrame.from_dict(VAR5, orient='index')
VAR1_df = pd.DataFrame.from_dict(VAR1, orient='index')

desc_stats['CV'] = desc_stats["Std"]/desc_stats["Mean"]
desc_stats['Beta'] = beta_df
# desc_stats['ADF test'] = ADF_test_df 
# desc_stats['ADF p-value'] = ADF_pval_df
# desc_stats['JB test'] = JB_test_df
# desc_stats['JB p-value'] = JB_pval_df
desc_stats['VAR 5%'] = VAR5_df
desc_stats['VAR 1%'] = VAR1_df

# desc_stats = desc_stats.round(5)
desc_stats['Name'] = returns.columns
# cols_to_select = desc_stats.columns.drop("Name")
(
    GT(desc_stats).cols_move_to_start(columns="Name")
    .tab_header(title="Descriptive Statistics for selected companies", 
    #    subtitle="S&P 500 used as refernence index"
    )
    .opt_stylize(5, color = "gray")
    .fmt(lambda x: f"{x:.2e}", columns=["Mean", "Std", 
    # "ADF p-value", "JB p-value"
    ])
    .fmt(lambda x: f"{x:.4f}", columns=["Skewness","Kurtosis", "CV", "Beta", 
    # "ADF test"
    ])
    .fmt(lambda x: f"{x:.4%}", columns=['VAR 5%', 'VAR 1%'])
    .fmt_integer(columns = "Count")


    # .fmt(lambda x: f"{x:.2f}", columns=["JB test"])

)
Descriptive Statistics for selected companies
Name Count Mean Std CV Skewness Kurtosis Beta VAR 5% VAR 1%
CVX US Equity 236 8.92e-04 1.89e-02 21.1958 -0.5396 4.2367 0.6900 -3.0219% -4.3108%
FSLR US Equity 236 1.89e-03 4.14e-02 21.9423 1.0953 7.0162 0.6614 -6.6172% -9.4370%
ENI IM Equity 236 8.19e-04 1.29e-02 15.7609 -1.0759 4.3374 0.0197 -2.0417% -2.9216%
XOM US Equity 236 4.38e-05 1.62e-02 370.2611 -0.8161 2.4833 0.6391 -2.6604% -3.7644%
SHEL LN Equity 236 5.07e-04 1.32e-02 26.0120 -1.0894 4.0888 0.1022 -2.1196% -3.0189%
VWS DC Equity 236 8.24e-04 3.65e-02 44.2567 0.4657 4.1393 -0.3037 -5.9134% -8.3976%
EQNR NO Equity 236 2.28e-04 2.06e-02 89.9712 -0.4305 1.9742 -0.0465 -3.3578% -4.7584%
TTE FP Equity 236 -1.61e-05 1.33e-02 -826.4150 -0.5504 1.5665 0.0221 -2.1924% -3.1001%
AMRC US Equity 236 2.69e-03 6.25e-02 23.2634 1.3758 19.2682 1.0578 -10.0197% -14.2824%
NEE US Equity 236 3.13e-04 1.81e-02 57.7545 -0.1578 1.5681 0.2967 -2.9461% -4.1797%
Show code 
low  = 1
mid  = 2
high = 3
avgstd = np.mean(desc_stats["Std"])
#| label: cumulative_performance
plt.figure(figsize=(14, 6))
for company in returns.columns:
    plt.plot(returns[company])
plt.legend(labels=returns.columns,loc='center left', bbox_to_anchor=(1.02, 0.5), 
    ncol=1,frameon=False)
plt.title("Daily discrete returns of companies last year", fontsize=16)
plt.xlabel("Date", fontsize=16)
# plt.axhline(low*avgstd, color='red', linestyle='--', label='Market Beta = 1')
# plt.axhline(mid*avgstd, color='red', linestyle='--', label='Market Beta = 1')
# plt.axhline(high*avgstd, color='red', linestyle='--', label='Market Beta = 1')
# plt.axhline(-low*avgstd, color='red', linestyle='--', label='Market Beta = 1')
# plt.axhline(-mid*avgstd, color='red', linestyle='--', label='Market Beta = 1')
# plt.axhline(-high*avgstd, color='red', linestyle='--', label='Market Beta = 1')
plt.ylabel("Discrete returns", fontsize=16)
plt.grid(True, linestyle='--', alpha=0.6)
plt.xticks(fontsize=14)  # shrink labels
plt.yticks(fontsize=14)  # shrink labels

plt.tight_layout()                 # makes space for title/axes
plt.subplots_adjust(right=0.8)     # extra: leaves space for legend
plt.show()

This chart allows us to spot outlier days. Ameresco had a big drop in price following an earnings call but then recovered with positive earnings later in the year (Omotosho 2025). First solar, First Solar also experienced a price spike in August following a US government announcement regarding tax credit for clean energy projects (Bohen 2025).

Show code 
plt.figure(figsize=(14, 6))
for company in returns.columns:
    plt.plot((np.cumsum(returns[company])))
plt.legend(labels=returns.columns,loc='center left', bbox_to_anchor=(1.02, 0.5), 
    ncol=1,frameon=False)
plt.title("Cumulative Performance of discrete returns", fontsize=16)
plt.xlabel("Date", fontsize=16)
plt.ylabel("Cumulative returns", fontsize=16)
plt.grid(True, linestyle='--', alpha=0.6)
plt.xticks(fontsize=14)  # shrink labels
plt.yticks(fontsize=14)  # shrink labels

plt.tight_layout()                 # makes space for title/axes
plt.subplots_adjust(right=0.8)     # extra: leaves space for legend
plt.show()

Show code 
plt.figure(figsize=(14, 6))
for company in df2.columns:
    plt.plot(df2[company])
plt.legend(labels=df2.columns,loc='center left', bbox_to_anchor=(1.02, 0.5), 
    ncol=1,frameon=False)
plt.title("Cumulative Performance of discrete returns", fontsize=16)
plt.xlabel("Date", fontsize=16)
plt.ylabel("Cumulative returns", fontsize=16)
plt.grid(True, linestyle='--', alpha=0.6)
plt.xticks(fontsize=14)  # shrink labels
plt.yticks(fontsize=14)  # shrink labels

plt.tight_layout()                 # makes space for title/axes
plt.subplots_adjust(right=0.8)     # extra: leaves space for legend
plt.show()

print("Day with the lowest returns")
returns.idxmin(axis=0).sort_values()

Day with the lowest returns
VWS DC Equity    2024-11-06
AMRC US Equity   2025-02-28
CVX US Equity    2025-04-03
SHEL LN Equity   2025-04-04
TTE FP Equity    2025-04-04
ENI IM Equity    2025-04-07
XOM US Equity    2025-04-10
FSLR US Equity   2025-06-17
EQNR NO Equity   2025-06-24
NEE US Equity    2025-07-23
dtype: datetime64[ns]

This chart allows us to spot outlier days. Ameresco had a big drop in price following an earnings call but then recovered with positive earnings later in the year (Omotosho 2025). First solar, First Solar also experienced a price spike in August following a US government announcement regarding tax credit for clean energy projects (Bohen 2025).

Show code 
cum_perf = (1 + returns).cumprod() - 1 +100

plt.figure(figsize=(14, 6))

for company in cum_perf.columns:
    plt.plot(cum_perf.index, cum_perf[company], label=company)

plt.legend(loc='center left', bbox_to_anchor=(1.02, 0.5),
           ncol=1, frameon=False)
plt.title("Cumulative Performance of Stocks", fontsize=16)
plt.xlabel("Date", fontsize=16)
plt.ylabel("Cumulative Return", fontsize=16)
plt.grid(True, linestyle='--', alpha=0.6)
plt.xticks(fontsize=14)
plt.yticks(fontsize=14)

plt.tight_layout()
plt.subplots_adjust(right=0.8)
plt.show()

This cumulative performance chart shows how all stocks performed starting from 100. Here, it is possible to see that stocks like, Chevron, Eni, Exxon, Shell, and TotalEnergies were hurt by the tariffs in April 2025. This was due to the implied increase in price for oil coming from overseas, which all these companies deal in, when tariffs raise fears of weaker global growth and reduced oil demand (International Energy Agency 2025). The renewables did not feel the same effects, proving their resilience to these kinds of geopolitical events.

Show code 
MKTL = MKT.iloc[3:,:]
MKTL[['CVX US Equity', 'XOM US Equity', 'SHEL LN Equity']]
# Compute group weights
GREEN_weights = MKTL[['VWS DC Equity', 'FSLR US Equity', 'AMRC US Equity']].div(
    MKTL[['VWS DC Equity', 'FSLR US Equity', 'AMRC US Equity']].sum(axis=1), axis=0
).iloc[0] 

TRANSITIONAL_weights = MKTL[['ENI IM Equity', 'TTE FP Equity', 'EQNR NO Equity', 'NEE US Equity']].div(
    MKTL[['ENI IM Equity', 'TTE FP Equity', 'EQNR NO Equity', 'NEE US Equity']].sum(axis=1), axis=0
).iloc[0] 

BROWN_weights = MKTL[['CVX US Equity', 'XOM US Equity', 'SHEL LN Equity']].div(
    MKTL[['CVX US Equity', 'XOM US Equity', 'SHEL LN Equity']].sum(axis=1), axis=0
).iloc[0] 


GREEN_ret = (returns[['VWS DC Equity', 'FSLR US Equity', 'AMRC US Equity']] * GREEN_weights).sum(axis=1)
TRANSITIONAL_ret = (returns[['ENI IM Equity', 'TTE FP Equity', 'EQNR NO Equity', 'NEE US Equity']] * TRANSITIONAL_weights).sum(axis=1)
BROWN_ret = (returns[['CVX US Equity', 'XOM US Equity', 'SHEL LN Equity']] * BROWN_weights).sum(axis=1)


cum_GREEN_ret = 100 * (1 + GREEN_ret).cumprod()
cum_TRANSITIONAL_ret = 100 * (1 + TRANSITIONAL_ret).cumprod()
cum_BROWN_ret = 100 * (1 + BROWN_ret).cumprod()

plt.figure(figsize=(12, 6))
plt.plot(cum_GREEN_ret, label='Aligned Companies', color='green')
plt.plot(cum_TRANSITIONAL_ret, label='Developing Companies', color='gray')
plt.plot(cum_BROWN_ret, label='Unprepared Companies', color='brown')

plt.grid(axis='both', linestyle='-', alpha=0.6)
plt.legend(title="Company", loc='center left', bbox_to_anchor=(1.02, 0.5), ncol=1, frameon=False)
plt.title("Market-Cap Weighted Cumulative Returns by ESG Category", fontsize=16)
plt.tight_layout()
plt.show()

GREEN = returns[['VWS DC Equity', 'FSLR US Equity', "AMRC US Equity"]].sum(axis=1) 
TRANSITIONAL = returns[['ENI IM Equity', 'TTE FP Equity', "EQNR NO Equity", "NEE US Equity"]].sum(axis=1) 
BROWN = returns[['CVX US Equity', 'XOM US Equity', "SHEL LN Equity"]].sum(axis=1) 

cum_GREEN_ret = 100 * (1 + GREEN).cumprod()
cum_TRANSITIONAL_ret = 100 * (1 + TRANSITIONAL).cumprod()
cum_BROWN_ret = 100 * (1 + BROWN).cumprod()


plt.figure(figsize=(12, 6)) 
plt.plot(cum_GREEN_ret, label='Aligned Companies', color='green') 
plt.plot(cum_TRANSITIONAL_ret, label='Developing Companies', color='gray') 
plt.plot(cum_BROWN_ret, label='Unprepared Companies', color='brown') 
plt.grid(axis='both', linestyle='-', alpha=0.6) 
plt.legend(title="Company", loc='center left', bbox_to_anchor=(1.02, 0.5), ncol=1, frameon=False) 
plt.title("Cumulative Returns by ESG Category Equally weighted", fontsize=16)

Text(0.5, 1.0, 'Cumulative Returns by ESG Category Equally weighted')

Just to get a final look at how these companies performed in the market this chart, this is a market cap weighted return chart where each daily return is multiplied by their market share among companies of the same climate readiness. The formula ends up being: w_{i,j}=\frac{m_{i,j}}{∑_{k∈j} m_{k,j}}

where i is the individual asset and j is the transition readiness and k means all assets i that belong to group j. With this, it is easy to see that although the climate-aligned companies collectively did not outperform the other two groups for most of the year, and in fact experienced a big drawdown for half of the year, they picked up pace and are now clear above the rest.

Show code 
import ptitprince as pt
melted = returns.melt(var_name="Asset", value_name="Return")

plt.figure(figsize=(14, 6))


pt.half_violinplot(
    x="Asset",
    y="Return",
    data=melted,
    palette="tab10",
    bw=.2,
    cut=0.,
    scale="area",
    inner=None,
    orient="v"
)
sns.boxplot(
    data=melted,
    x="Asset",
    y="Return",
    width=0.2,
    palette="tab10",
    showcaps=True,
    flierprops = dict(marker='o', markerfacecolor='black', markersize=2, linestyle='none')
    # showfliers=False
)


# 1. Horizontal Line at y=0 (Zero Return)
plt.axhline(
    y=0, 
    color='black', 
    linestyle='--', 
    linewidth=0.8, 
    alpha=0.7,
    label='Zero Return' # Adding a label is good practice
)

# 2. Vertical Lines to separate Assets/Boxplots
for i in range(9): # 10 - 1
    plt.axvline(
        x=i + 0.5, # Place the line exactly between asset i and asset i+1
        color='gray', 
        linestyle='-', 
        linewidth=0.5, 
        alpha=0.5
    )


plt.title("Violin + Boxplot of Asset Returns")
plt.xticks(rotation=45)
plt.tight_layout()
plt.show()

This violin and boxplots show the distribution of the companies’ returns and demonstrate that Firms such as First Solar, Vestas and Ameresco indicate a more spread distribution, showing higher risks, while TotalEnergies and Eni indicate more stable returns.

Correlation Analysis

Show code 
corr_matrix = returns.corr(method = 'pearson')
plt.figure(figsize=(10, 8))
sns.heatmap(corr_matrix, annot=True, cmap='coolwarm', fmt=".2f", annot_kws={"size": 10})
plt.title("Correlation Matrix of Returns")
plt.tight_layout()

The correlation plot shows clearly the correlation that ENI, Exxon, Chevron, Shell, Equinor and TotalEnergies still have with each other, mainly due to their connection to the price of oil, while the rest do not really show strong correlations due to their businesses being overall diversified, so sector-specific changes did not impact them equally at the same time. Lastly, there is also no negative correlation to the oil giants; this might be a symptom of the overall difference in volatility experienced in the market.

Show code 
plt.figure(figsize=(10, 8))
sns.heatmap(returns.corr(method = 'spearman') - returns.corr(method = 'pearson'), annot=True, cmap='coolwarm', fmt=".2f", annot_kws={"size": 10})
plt.title("Difference in correlation Matrix of Returns")
Text(0.5, 1.0, 'Difference in correlation Matrix of Returns')

Financial Metrics

Show code 
PE_long  =  PE.set_index('Date').stack().reset_index().rename(columns = {"level_1":"Company", 0:"PE"})
MKT_long = MKT.set_index('Date').stack().reset_index().rename(columns = {"level_1":"Company", 0:"MKT"})
EPS_long = EPS.set_index('Date').stack().reset_index().rename(columns = {"level_1":"Company", 0:"EPS"})
ROA_long = ROA.set_index('Date').stack().reset_index().rename(columns = {"level_1":"Company", 0:"ROA"})
ROE_long = ROE.set_index('Date').stack().reset_index().rename(columns = {"level_1":"Company", 0:"ROE"})
ESG_long = ESG.set_index('Date').stack().reset_index().rename(columns = {"level_1":"Company", 0:"ESG"})

df_panel = pd.merge(PE_long,  MKT_long, on=['Date', 'Company'])
df_panel = pd.merge(df_panel, EPS_long, on=['Date', 'Company'])
GT(df_panel.sort_values(['Company']))
Date Company PE MKT EPS
Q1 2025 AMRC US Equity 18.6753 491.747 -0.0794
Q2 2025 AMRC US Equity 18.9435 582.6516 0.1798
Q3 2025 AMRC US Equity 40.7588 1314.9483 0.2596
Q4 2024 AMRC US Equity 35.7656 983.5896 0.5543
Q4 2024 CVX US Equity 17.5221 207916.5329 1.4443
Q3 2025 CVX US Equity 27.0581 236228.397 1.3572
Q1 2025 CVX US Equity 23.4966 226783.8792 1.596
Q2 2025 CVX US Equity 23.6984 182343.695 1.0861
Q3 2025 ENI IM Equity 18.3212 40864.1228 0.2166
Q1 2025 ENI IM Equity 18.7783 37689.4974 0.3008
Q4 2024 ENI IM Equity 16.9689 35622.1154 0.0345
Q2 2025 ENI IM Equity 19.0738 37139.5564 0.1359
Q3 2025 EQNR NO Equity 11.5145 46373.2035 -0.0593
Q2 2025 EQNR NO Equity 8.3402 51462.8834 0.3745
Q1 2025 EQNR NO Equity 8.331 57323.5871 0.7702
Q4 2024 EQNR NO Equity 7.4988 52084.9723 0.5699
Q3 2025 FSLR US Equity 16.0091 17572.8691 3.1519
Q1 2025 FSLR US Equity 10.181 10487.2255 1.5563
Q2 2025 FSLR US Equity 13.3798 12945.3486 2.3894
Q4 2024 FSLR US Equity 13.9387 15071.3753 2.8651
Q2 2025 NEE US Equity 18.6761 104207.5043 0.7415
Q1 2025 NEE US Equity 19.8611 112979.4631 0.3256
Q3 2025 NEE US Equity 19.6291 115503.3788 0.8751
Q4 2024 NEE US Equity 22.2759 117759.9181 0.4528
Q2 2025 SHEL LN Equity 15.6333 150971.7067 0.4569
Q3 2025 SHEL LN Equity 14.4794 153793.5592 0.6749
Q1 2025 SHEL LN Equity 16.7258 169983.4394 0.6273
Q4 2024 SHEL LN Equity 12.2518 151408.1715 0.1171
Q3 2025 TTE FP Equity 9.7443 99711.8954 1.2237
Q1 2025 TTE FP Equity 10.8453 113449.0273 1.3419
Q4 2024 TTE FP Equity 8.2712 105783.2996 1.3439
Q2 2025 TTE FP Equity 11.0459 101444.493 0.8839
Q4 2024 VWS DC Equity 26.2158 11004.1938 0.5008
Q2 2025 VWS DC Equity 16.7213 11032.0113 0.0255
Q3 2025 VWS DC Equity 17.2261 14157.8123 0.26
Q1 2025 VWS DC Equity 22.2552 10762.73 0.0
Q3 2025 XOM US Equity 16.7982 357144.7371 1.3053
Q4 2024 XOM US Equity 13.8069 377650.227 1.3428
Q2 2025 XOM US Equity 15.4197 338762.6257 1.2284
Q1 2025 XOM US Equity 15.8763 399825.1447 1.3975
Show code 
df_panel_ratios = pd.merge(ROE_long, ROA_long, on=['Date', 'Company'])
df_panel_ratios = pd.merge(df_panel_ratios, ESG_long, on=['Date', 'Company'])
GT(df_panel_ratios)
Date Company ROE ROA ESG
Q3 2025 CVX US Equity 7.3794 4.3597 3.03
Q3 2025 FSLR US Equity 16.8639 11.2505 5.57
Q3 2025 ENI IM Equity 5.2792 2.0033 7.09
Q3 2025 XOM US Equity 18.3731 6.5381 3.27
Q3 2025 SHEL LN Equity 8.0496 3.7868 3.51
Q3 2025 VWS DC Equity 29.4007 3.8371 5.11
Q3 2025 EQNR NO Equity 13.4925 4.2264 2.19
Q3 2025 TTE FP Equity 12.2564 4.9229 6.53
Q3 2025 AMRC US Equity 6.3219 1.4984 1.68
Q3 2025 NEE US Equity 12.476 3.3312 6.56

Scatter plots

Show code 
Quick_scatter(EPS, PE, "Earnings per share", "PE ratio")

The P/E and EPS are two ratios that are also connected in an inverse relation as .investors understand that this has been a difficult year for the sector and are betting on the results being better next year, increasing the P/E when the EPS is lower and vice versa, this is called the Molodovsky effect and although it is not present in all industries, in cyclical ones like the energy one it can be observed and used to explain such behaviour (Corporate Finance Institute 2024). This appears to be more present in the greener companies than in the oil companies, as they are seen as companies with more potential.

Show code 
Quick_scatter(PE, MKT, "PE ratio", "Market cap")

The relationship between Earnings per Share (EPS) and Market Capitalization is presented showing weak investor sentiment towards larger companies, usually smaller companies have more hype and expectations behind them like Vestas and Ameresco, which are the 2 smallest firms by market capitalisation, but show higher P/E compared to giants like Exxon that for their massive market capitalisation of 4.2 billion trades at a valuation lower than expected for its size, alongside Chevron these more established companies are in the later growth and are thus being priced for reality not for dreams.

Show code 
def stacked_bar_chart(df, name):
    quarters = ["Q4 2024", "Q1 2025", "Q2 2025", "Q3 2025"]
    plot_df = df.transpose().iloc[-10:]
    plot_df.columns = quarters
    companies = plot_df.index.tolist()

    # Convert all values to float
    plot_df = plot_df.apply(pd.to_numeric, errors="coerce")

    x = np.arange(len(quarters))

    plt.figure(figsize=(14,7))

    bottom = np.zeros(len(quarters), dtype=float)

    for company in companies:
    
        values = plot_df.loc[company, quarters].values.astype(float)

        plt.bar(
            x,
            values,
            bottom=bottom,
            label=company
        )
        bottom += values  # numeric safe

    plt.xticks(x, quarters)
    plt.xlabel("Quarter")
    plt.ylabel(f"Total Value of {name} (Stacked)")
    plt.title(f"Stacked Bar Chart of {name} (per Quarter)")
    plt.legend(title="Company", loc='center left', bbox_to_anchor=(1.02, 0.5), frameon=False)
    plt.tight_layout()
    plt.show()


stacked_bar_chart(MKT, "Market Capitalisation")

The stacked bar chart allows to see that the market overall did not expand that much over the last year; the overall market capitalisation of these firms increased only by 3.97% from 2024, mainly due to the hardships faced in the second quarter with tariffs and oil price pressure.

ESG analysis

Show code 
MKT2 = MKT.iloc[3:,:]
PE2 = PE.iloc[3:,:]
EPS2 = EPS.iloc[3:,:]

Quick_scatter(ROE, ESG, "Returns on Equity", "ESG score")
Quick_scatter(ROA, ESG, "Returns on Assets", "ESG score")
Quick_scatter(MKT2, ESG, "Market Capitalisation", "ESG score")

Show code 
Quick_scatter(ROE, ESG, "Returns on Equity", "ESG score", panelreg = True)
Quick_scatter(ROA, ESG, "Returns on Assets", "ESG score", panelreg = True)
Quick_scatter(PE2, ESG, "Price Earnings", "ESG score", panelreg = True)
Quick_scatter(EPS2, ESG, "Earnings per Share", "ESG score", panelreg = True)

When looking at the ESG score, the story is clear: the market still wants companies to move towards greener investments. Across our sample, clean companies are consistently associated with stronger financial performance. While ROA highlights the operational efficiency of the asset-heavy oil majors, the ROE is crucial for understanding the renewable firms, which rely heavily on equity financing for growth. Interestingly, firms with better sustainability profiles tended to outperform in both metrics, suggesting that the ‘green premium’ is supported by genuine operational efficiency and lower risk premia. This relationship extends to the EPS as well: companies with stronger ESG credentials typically command larger revenues over the last year. Overall, the evidence points to a positive correlation between ESG quality and both profitability and market perception in all but the P/E ratio, as higher ESG scores can signal higher costs and lower short-term profitability, which can pressure P/E ratios. Meanwhile, Oil companies show no correlation to any ESG variables, implying that investors and other market participants do not value them based on ESG.

Conclusion

This study provides empirical evidence of a structural divergence within the energy sector, driven by the ongoing global energy transition. The analysis of financial metrics and descriptive statistics confirms that the sector is split between established incumbents acting as defensive anchors and volatile disruptors representing speculative growth. These companies have stood the test of a rather difficult year and have also shown resilience, especially when it comes to the looming threat of tariffs and different kinds of market pressures that have impacted oil and gas in the last year, whether it is inflationary, geopolitical tensions, including armed conflicts and trade restrictions. The renewable segment exhibits higher probability of extreme events compared to the integrated oil majors. This volatility is further supported by Beta coefficients; firms such as Ameresco exceed market risk, while integrated Oil companies like Shell and Chevron act as defensive assets. The scatter plot analysis of fundamentals reveals a market preference for “growth potential” over “current value”, with investors that appear willing to assign higher valuation multiples to firms despite lower or volatile earnings, banking on future growth. Contrary to standard liquidity premiums, the analysis suggests that larger capitalised firms (i.e., Exxon, Shell) trade at lower P/E multiples. This indicates that the market prices these giants for “reality”, discounting their mature growth profiles, while actively pricing smaller renewables for “dreams,” with a hype premium. Finally, the ESG analysis highlights a maturing market rationality. There is a consistent positive correlation between ESG scores and profitability metrics. This suggests that sustainable practices are no longer just a compliance cost but are associated with operational efficiency. However, the lower P/E ratios often found in the highest ESG scorers imply that investors remain cautious about the capital intensity required for the transition, especially in the short term.

References

Bohen, Tim. 2025. “FSLR Stock Surge: What’s Next?” Stockstotrade, August 15, 2025. https://stockstotrade.com/news/first-solar-inc-fslr-news-2025_08_15-2

Corporate Finance Institute. 2024. “Molodovsky Effect.” Corporate Finance Institute, September 10, 2024. https://corporatefinanceinstitute.com/resources/career-map/sell-side/capital-markets/molodovsky-effect International Energy Agency. 2025. “Oil Market Report for April 2025,” April 2025. https://www.iea.org/reports/oil-market-report-april-2025. Omotosho, Kayode. 2025. “Why Ameresco (AMRC) Stock Is Trading Lower Today.” Yahoo Finance, February 28, 2025. https://finance.yahoo.com/news/why-ameresco-amrc-stock-trading-192128114.html.

Appendix

Show code 
Quick_scatter(ROE, PE2,  "ROE", "PE", panelreg = False, groupreg = False)
# Horizontal and vertical thresholds
x_thresh = 7    # ROE
y_thresh = 30   # PE

# Shade region: x < 10 (vertical span), y > 30 (horizontal span)
plt.axvspan(xmin=plt.xlim()[0], xmax=x_thresh, ymin=y_thresh / plt.ylim()[1], ymax=1,
color='red', alpha=0.15, label='Low ROE & High PE Region')

x_thresh = 10   # ROE
y_thresh = 15   # PE
# Shade region: x < 10 (vertical span), y > 30 (horizontal span)
plt.axvspan(xmin=plt.xlim()[0], xmax=x_thresh, ymin=y_thresh / plt.ylim()[1], ymax=1,
color='red', alpha=0.15, label='Low ROE & High PE Region')

Show code 
def Quick_scatter(Data1, Data2, label1, label2, ylog=False, xlog=False, panelreg = False, groupreg = True):
        # Collect all data for panel regression
    all_data1 = []
    all_data2 = []
    plt.figure(figsize=(8, 6))
    
    if groupreg:
    # ---- classify companies ----
        green_status = {
            "CVX US Equity": "no",
            "FSLR US Equity": "yes",
            "ENI IM Equity": "some",
            "XOM US Equity": "no",
            "SHEL LN Equity": "no",
            "VWS DC Equity": "yes",
            "EQNR NO Equity": "some",
            "TTE FP Equity": "some",
            "AMRC US Equity": "yes",
            "NEE US Equity": "yes"
        }


        # If you want: define green = yes only
        green_companies = [c for c,s in green_status.items() if s == "yes"]
        non_green_companies = [c for c,s in green_status.items() if s == "no"]
        some_companies = [c for c,s in green_status.items() if s == "some"]  # optional


        # Also prepare lists for green/non-green group regressions
        green_x = []
        green_y = []
        nongreen_x = []
        nongreen_y = []
        some_x = []
        some_y = []

    # --- Plot scatter + company-level regressions ---
    for company in company_columns:
        Data1_series = Data1[company]
        Data2_series = Data2[company]

        all_data1.append(Data1_series)
        all_data2.append(Data2_series)
        if groupreg:
            # Add to correct group
            if company in green_companies:
                green_x.append(Data1_series)
                green_y.append(Data2_series)
            elif company in non_green_companies:
                nongreen_x.append(Data1_series)
                nongreen_y.append(Data2_series)
            elif company in some_companies:
                some_x.append(Data1_series)
                some_y.append(Data2_series)

        # Company scatter
        plt.scatter(Data1_series, Data2_series, label=company)

        # Company individual regression
        # sns.regplot(
        #     x=Data1_series.values,
        #     y=Data2_series.values,
        #     scatter=False,
        #     ci=None,
        #     line_kws={'alpha': 0.6, 'linewidth': 1}
        # )

    # --- Convert group lists ---
    if groupreg:
        gx = pd.concat(green_x).values if green_x else None
        gy = pd.concat(green_y).values if green_y else None
        nx = pd.concat(nongreen_x).values if nongreen_x else None
        ny = pd.concat(nongreen_y).values if nongreen_y else None
        sx = pd.concat(some_x).values if some_x else None
        sy = pd.concat(some_y).values if some_y else None

    if panelreg:
        # --- Panel-wide regression ---
        all_x = pd.concat(all_data1).values
        all_y = pd.concat(all_data2).values
        sns.regplot(
            x=all_x, y=all_y,
            scatter=False, ci=None,
            label='Panel-Wide Regression',
            line_kws={'color': 'blue', 'linewidth': 4, 'alpha': 0.2}
        )

    if groupreg:
        # --- Group Regressions ---
        if gx is not None:
            sns.regplot(
                x=gx, y=gy,
                scatter=False, ci=None,
                label='Aligned Companies',
                line_kws={'color': 'green', 'linewidth': 5, 'alpha': 0.2}
            )

        if nx is not None:
            sns.regplot(
                x=nx, y=ny,
                scatter=False, ci=None,
                label='Unprepared Companies',
                line_kws={'color': 'red', 'linewidth': 5, 'alpha': 0.2}
            )

        if sx is not None:
            sns.regplot(
                x=sx, y=sy,
                scatter=False, ci=None,
                label='Developing Companies',
                line_kws={'color': 'gray', 'linewidth': 5, 'alpha': 0.2}
        )

    # --- Formatting ---
    if xlog:
        plt.xscale('log')
    if ylog:
        plt.yscale('log')

    plt.title(f'{label1} vs. {label2}', fontsize=14)
    plt.xlabel(label1)
    plt.ylabel(label2)
    plt.grid(True, ls="--", alpha=0.6)
    plt.legend(loc='center left', bbox_to_anchor=(1.02, 0.5), frameon=False)
    plt.tight_layout()

Quick_scatter(ROE, ESG, "Returns on Equity", "ESG score")
Quick_scatter(ROA, ESG, "Returns on Assets", "ESG score")
Quick_scatter(PE2, ESG, "Price Earnings", "ESG score")
Quick_scatter(EPS2, ESG, "Earnings per Share", "ESG score")

Show code 
import matplotlib.pyplot as plt
# Assuming 'returns' DataFrame is already defined, and 'window' is set to 30

window = 30
rolling_mean = returns.rolling(window=window).mean()
rolling_vol = returns.rolling(window=window).std()

# --- Chart 1: Rolling Mean ---
plt.figure(figsize=(12, 6))

for col in returns.columns:
    # Plotting each asset's rolling mean
    plt.plot(rolling_mean.index, rolling_mean[col], linewidth=1.5, label=col)

plt.title(f'{window}-Day Rolling Mean of Returns', fontsize=14)
plt.xlabel('Date')
plt.ylabel('Rolling Mean Return')
plt.grid(True, linestyle='--', alpha=0.6)

# Legend for Chart 1
plt.legend(
    title='Asset',
    loc='center left',
    bbox_to_anchor=(1.02, 0.5), # Moves the legend outside the plot area
    frameon=False # Ensure a background for the legend for clarity
)
plt.tight_layout() # Adjust layout to make room for the legend
plt.show()


# --- Chart 2: Rolling Volatility ---
plt.figure(figsize=(12, 6))

for col in returns.columns:
    # Plotting each asset's rolling volatility
    plt.plot(rolling_vol.index, rolling_vol[col], linewidth=1.5, label=col)

plt.title(f'{window}-Day Rolling Volatility (Standard Deviation)', fontsize=14)
plt.xlabel('Date')
plt.ylabel('Rolling Volatility (Std Dev)')
plt.grid(True, linestyle='--', alpha=0.6)

# Legend for Chart 2 (identical placement)
plt.legend(
    title='Asset',
    loc='center left',
    bbox_to_anchor=(1.02, 0.5),
    frameon=False
)
plt.tight_layout()
plt.show()

Show code 
sorted_beta = beta_df.sort_values('Beta', ascending=False)
mycolors = ['red' if v < 0 else 'steelblue' for v in sorted_beta['Beta']]
ax = sorted_beta.plot(kind='bar', color=mycolors, legend=False, figsize=(9, 6))
plt.axhline(1, color='red', linestyle='--', label='Market Beta = 1')
plt.title('Market Beta per Company')
plt.ylabel('Beta')
plt.legend()
plt.tight_layout()
plt.show()

Show code 
basic_plot(PE, "P/E ratio")
basic_plot(MKT, "market capitalisation")
basic_plot(EPS, "Earnings per share")
# basic_plot(ROA, "return on assets")
# basic_plot(ROE, "return on equity")
# basic_plot(ESG, "Overall esg score")

Show code 
clustered_bar_chart(MKT, "Market Capitalisation")
clustered_bar_chart(EPS, "Earnings Per Share")
clustered_bar_chart(PE, "P/E Ratio")

Scatter with groups

Show code 
Quick_scatter(PE, MKT, "PE ratio", "Market cap", ylog = False, panelreg = True)
Quick_scatter(EPS, MKT, "Earnings per share", "Market cap", ylog = False, panelreg = True)
Quick_scatter(EPS, PE, "Earnings per share", "PE ratio", panelreg = True)
Quick_scatter(ROE, ROA, "Returns on Equity", "Returns on Assets", panelreg = True)
# Quick_scatter_1reg(ESG, PE, "ESG score", "PE ratio")
# Quick_scatter_1reg(ESG, EPS, "ESG score", "Earnings per share")
# Quick_scatter_1reg(ESG, ROE, "ESG score", "Returns on Equity")
# Quick_scatter_1reg(ESG, ROA, "ESG score", "Returns on Assets")

Show code 
Quick_scatter(EPS, MKT, "Earnings per share", "Market cap")
Quick_scatter(ROE, ROA, "Returns on Equity", "Returns on Assets")

Show code 
stacked_bar_chart(PE, "P/E Ratio")
stacked_bar_chart(EPS, "Earnings Per Share")

Show code 
percentage_stacked_bar_chart(MKT, "Market Capitalisation")
percentage_stacked_bar_chart(EPS, "Earnings Per Share")
percentage_stacked_bar_chart(PE, "P/E Ratio")

Show code 
Quick_scatter(MKT2, ESG, "Market Capitalisation", "ESG score", panelreg = True)

Growth

Show code 
basic_plot(EPS.iloc[:, 1:].div(EPS.iloc[0, 1:], axis=1), "Earnings per Share Growth")
basic_plot(MKT.iloc[:, 1:].div(MKT.iloc[0, 1:], axis=1), "Market Capitalisation Growth")
basic_plot(PE.iloc[:, 1:].div(PE.iloc[0, 1:], axis=1), "PE Growth")

Show code 
simple_bar_chart(ROA, "Return on Assets (%)")
simple_bar_chart(ROE, "Return on Equity (%)")
simple_bar_chart(ESG, "ESG Scores")
DIFF = (ROE.transpose().iloc[-10:] - ROA.transpose().iloc[-10:])
simple_bar_chart(DIFF.transpose(), "ROE - ROA")